This repository contains the code to reproduce the simulation experiments in:
Ponge, J., Patzner, J., Hellingrath, and Karch, A., 2025. Targeted Household Quarantining: Enhancing the Efficiency of Epidemic Responses. Submitted to 2025 Winter Simulation Conference (WSC)
Non-pharmaceutical interventions (NPIs) are the immediate public health reaction to emerging epidemics. While they generally help slow down infection dynamics, they can be associated with relevant socioeconomic costs, like lost school- or work days caused by preemptive household quarantines. However, research suggests that not all households contribute equally to the overall infection dynamics. In this study, we introduce the novel “Infection Contribution” metric that allows us to trace the involvement of particular household types over entire infection chains. Building upon the German Epidemic Microsimulation System, we quantify the impact of various household types, considering their size and composition in a COVID-1 6FB3 9- like scenario. Additionally, we show how targeting interventions based on household characteristics produces efficient strategies, outperforming non-selective strategies in almost all scenarios. Our approach can be transferred to other NPIs, such as school closure, testing, or contact tracing, and even inform the prioritization of vaccinations.
Our agent-based simulations are done in v0.4.3 of the German Epidemic Microsimulation System (GEMS) Julia package. We use the population model for the Western-German federal state of Saarland with roughly one million individuals.
Moreover, we simulate a COVID-19-like disease scenario to test various intervention strategies. Here's what an unmitigated case looks like:
The graphs below show how quarantining strategies targeting households based on various composition- and size-attributes differ significantly regarding their potential to "flatten the curve" (R0-Reduction) and the incurred quarantine days as well as lost school- and workdays.
Given that you have Julia 1.11.3+ installed,
clone the repository and run pkg> instantiate in your repository clone to download all dependencies.
You can set up a new script (e.g., experiments.jl, which is already on .gitignore) in your root folder to work with.
Then, load the package and run the experiments like this:
using GEMS_HH_Isolations
run_experiments()This will generate a new folder in your current working directory called results.
Every time you execute run_experiments(), a new timestamped subfolder will be generated in your results folder that will contain raw data outputs, processed and summarized data about the results, and the graphs that you see above (among others).
The default number of simulation runs for the baseline scenario and each of the quarantining scenarios is 10 which will take multiple hours on most machines to complete given that there are 72 scenarios.
You can pass the optional num_of_baseline_sims and num_of_scenario_sims arguments to decrease (or increase) the number of simulation runs.
If you want to re-run these experiments with other configurations (some options are in the /models folder), pass the input_config argument with the path to a configfile.
To change the baseline population, pass the population_model arugument with a German state identifier (e.g., Saarland is SL; Berlin is BE; Sachsen-Anhalt is ST; ...)